400 Gigabit Ethernet Task Force ready to get to work

March 28, 2014
The effort to create specifications for 400 Gigabit Ethernet (400GbE) has reached a major milestone with the formal creation of the IEEE P802.3bs Task Force. John D’Ambrosia, chair of the 400GbE Study Group within the IEEE (see “IEEE 802.3 400 Gigabit Ethernet Study Group chair warms to task”) as well as chair of the Ethernet Alliance, was named acting chair of the Task Force. D’Ambrosia, who should be formally installed as Task Force chairman at the IEEE 802.3/IEEE 802.1 Joint Interim Meeting in Norfolk, VA, the week of May 12, 2014, faces a very full plate of decisions and issues that must be cleared in the path toward a ratified standard.

The effort to create specifications for 400 Gigabit Ethernet (400GbE) has reached a major milestone with the formal creation of the IEEE P802.3bs Task Force. John D’Ambrosia, chair of the 400GbE Study Group within the IEEE (see “IEEE 802.3 400 Gigabit Ethernet Study Group chair warms to task”) as well as chair of the Ethernet Alliance, was named acting chair of the Task Force. D’Ambrosia, who should be formally installed as Task Force chairman at the IEEE 802.3/IEEE 802.1 Joint Interim Meeting in Norfolk, VA, the week of May 12, 2014, faces a very full plate of decisions and issues that must be cleared in the path toward a ratified standard.

The Task Force will tackle Ethernet Media Access Control (MAC) parameters, physical layer specifications, and management parameters for Ethernet transmission at 400 Gbps. This will include the creation of Physical Medium Dependent Sublayers (PMDs) for such applications as 100 m over multimode fiber and 500 m, 2 km, and 10 km via singlemode fiber.

D’Ambrosia noted that several technological issues will need to be straightened out to achieve these goals. Among the most salient are:

  • Sorting among approaches based on the use of multiple wavelengths, multiple fibers, and/or such modulation formats as discrete multi-tone (DMT) and pulse-amplitude modulation (PAM) for each PMD.
  • Establishing an architecture framework that will support interoperability among what D’Ambrosia expects will be multiple generations of 400GbE. For example, the creation of an approach using 16 lanes of 25 Gbps can be expected, but D’Ambrosia predicts that there will be interest in establishing specifications for 8x50-Gbps and 4x100-Gbps flavors as well. If one looks at a technology such as forward error correction (FEC), how should the use of FEC be specified, possibly for both the electrical lanes as well as the optical ones, to minimize incompatibilities among the different generations?
  • Developing the necessary electrical interfaces for chip-to-chip and chip-to-module communications.

It’s too early to predict how all of this will sort out. But D’Ambrosia said he expects the multimode PMD will use a parallel fiber approach. A parallel approach also may be appealing for the 500-m singlemode PMD, particularly to support 100GbE breakouts. However, D’Ambrosia believes that, based on reaction to the IEEE 802.3bm 40 and 100 Gigabit Ethernet specifications, users prefer single-fiber approaches to longer-reach applications such as 2 and 10 km.

As far as how many wavelengths each PMD might require, D’Ambrosia notes that “narrower is always better,” which is why the 100GbE community is moving towards 4x25G approaches. However, there currently isn’t an economical way to support the serial transmission of 100GbE that a 4x100-Gbps approach would require (although the Ethernet Alliance and the OIDA will co-host a workshop this June in San Jose on the subject). On the other hand, an approach that uses 16 wavelengths of 25 Gbps to accommodate 400 Gbps in a single fiber is likely too inelegant to pass muster, D’Ambrosia believes. For this reason, he expects there to be discussion of a four-wavelength-per-fiber approach for parallel fiber and an attempt to limit the number of lambdas to no more than eight for the single-fiber PMDs.

The Task Force also will work with the ITU-T to promote compatibility with that organization’s Optical Transport Network specifications (although there isn’t a 400G OTN specification yet).

If one looks at previous Gigabit Ethernet specification efforts as a model, the 400GbE standard could be up for ratification in 2017. D’Ambrosia acknowledged this, but noted that the time required to reach consensus on all the issues facing the Task Force is difficult to predict.

Meanwhile, the Ethernet Alliance has formed a 400GbE Subcommittee to augment the IEEE’s efforts. Several member companies of the Alliance will have employees active in the P802.3bs specifications work, and the subcommittee will provide a forum to build consensus around various proposals that would carry over to the Task Force, D’Ambrosia said. It also will build awareness of 400GbE. The subcommittee plans a member-only meeting on April 29, 2014.

For more information on optical transceivers and suppliers, visit the Lightwave Buyer’s Guide.